def test_get_parametric_dataset(self):
error_script1 = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
"""
error_script2 = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
"""
script = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
j.somethingelse('other')
"""
self.assertRaises(BackendError, get_parametric_datasets,
error_script1.splitlines())
self.assertRaises(BackendError, get_parametric_datasets,
error_script2.splitlines())
self.assertEqual(get_parametric_datasets(script.splitlines()),
[['a', 'b', 'c'], ['d', 'e', 'f'], ['g', 'h', 'i']],
"parametric dataset not correctly extracted")
self.assertTrue(
isinstance(get_parametric_datasets(script.splitlines()), list))
def test_get_parametric_dataset(self):
error_script1 = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
"""
error_script2 = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
"""
script = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
j.somethingelse('other')
"""
self.assertRaises(
BackendError, get_parametric_datasets, error_script1.splitlines())
self.assertRaises(
BackendError, get_parametric_datasets, error_script2.splitlines())
self.assertEqual(get_parametric_datasets(script.splitlines()),
[['a', 'b', 'c'], ['d', 'e', 'f'], ['g', 'h', 'i']],
"parametric dataset not correctly extracted")
self.assertTrue(
isinstance(get_parametric_datasets(script.splitlines()), list))
def _setup_bulk_subjobs(self, dirac_ids, dirac_script):
"""
This is the old bulk submit method which is used to construct the subjobs for a parametric job
Args:
dirac_ids (list): This is a list of the Dirac ids which have been created
dirac_script (str): Name of the dirac script which contains the job jdl
"""
f = open(dirac_script, 'r')
parametric_datasets = get_parametric_datasets(f.read().split('\n'))
f.close()
if len(parametric_datasets) != len(dirac_ids):
raise BackendError(
'Dirac',
'Missmatch between number of datasets defines in dirac API script and those returned by DIRAC'
)
master_job = self.getJobObject()
master_job.subjobs = []
for i in range(len(dirac_ids)):
j = Job()
j.copyFrom(master_job)
j.splitter = None
j.backend.id = dirac_ids[i]
j.id = i
j.inputdata = self._setup_subjob_dataset(parametric_datasets[i])
j.status = 'submitted'
j.time.timenow('submitted')
master_job.subjobs.append(j)
return True
def _setup_bulk_subjobs(self, dirac_ids, dirac_script):
"""
This is the old bulk submit method which is used to construct the subjobs for a parametric job
Args:
dirac_ids (list): This is a list of the Dirac ids which have been created
dirac_script (str): Name of the dirac script which contains the job jdl
"""
f = open(dirac_script, 'r')
parametric_datasets = get_parametric_datasets(f.read().split('\n'))
f.close()
if len(parametric_datasets) != len(dirac_ids):
raise BackendError('Dirac', 'Missmatch between number of datasets defines in dirac API script and those returned by DIRAC')
from Ganga.GPIDev.Lib.Job.Job import Job
master_job = self.getJobObject()
master_job.subjobs = []
for i in range(len(dirac_ids)):
j = Job()
j.copyFrom(master_job)
j.splitter = None
j.backend.id = dirac_ids[i]
j.id = i
j.inputdata = self._setup_subjob_dataset(parametric_datasets[i])
j.status = 'submitted'
j.time.timenow('submitted')
master_job.subjobs.append(j)
return True
def test_get_parametric_dataset():
error_script1 = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
"""
error_script2 = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
"""
script = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
j.somethingelse('other')
"""
with pytest.raises(BackendError):
get_parametric_datasets(error_script1.splitlines())
with pytest.raises(BackendError):
get_parametric_datasets(error_script2.splitlines())
assert get_parametric_datasets(script.splitlines()) == [
['a', 'b', 'c'], ['d', 'e', 'f'], ['g', 'h', 'i']
], 'parametric dataset not correctly extracted'
assert isinstance(get_parametric_datasets(script.splitlines()), list)
def test_get_parametric_dataset():
error_script1 = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
"""
error_script2 = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
"""
script = """
from DIRAC import Job
j=Job()
j.outputsomething('output.root')
j.setParametricInputData([['a','b','c'],['d','e','f'],['g','h','i']])
j.somethingelse('other')
"""
with pytest.raises(BackendError):
get_parametric_datasets(error_script1.splitlines())
with pytest.raises(BackendError):
get_parametric_datasets(error_script2.splitlines())
assert get_parametric_datasets(script.splitlines()) == [['a', 'b', 'c'], ['d', 'e', 'f'], ['g', 'h', 'i']], 'parametric dataset not correctly extracted'
assert isinstance(get_parametric_datasets(script.splitlines()), list)
def _setup_bulk_subjobs(self, dirac_ids, dirac_script):
f = open(dirac_script, 'r')
parametric_datasets = get_parametric_datasets(f.read().split('\n'))
f.close()
if len(parametric_datasets) != len(dirac_ids):
raise BackendError('Dirac', 'Missmatch between number of datasets defines in dirac API script and those returned by DIRAC')
from Ganga.GPIDev.Lib.Job.Job import Job
master_job = self.getJobObject()
for i in range(len(dirac_ids)):
j = Job()
j.copyFrom(master_job)
j.splitter = None
j.backend.id = dirac_ids[i]
j.id = i
j.inputdata = self._setup_subjob_dataset(parametric_datasets[i])
j.status = 'submitted'
j.time.timenow('submitted')
master_job.subjobs.append(j)
master_job._commit()
return True
def _setup_bulk_subjobs(self, dirac_ids, dirac_script):
f = open(dirac_script, 'r')
parametric_datasets = get_parametric_datasets(f.read().split('\n'))
f.close()
if len(parametric_datasets) != len(dirac_ids):
raise BackendError('Dirac', 'Missmatch between number of datasets defines in dirac API script and those returned by DIRAC')
from Ganga.GPIDev.Lib.Job.Job import Job
master_job = self.getJobObject()
master_job.subjobs = []
for i in range(len(dirac_ids)):
j = Job()
j.copyFrom(master_job)
j.splitter = None
j.backend.id = dirac_ids[i]
j.id = i
j.inputdata = self._setup_subjob_dataset(parametric_datasets[i])
j.status = 'submitted'
j.time.timenow('submitted')
master_job.subjobs.append(j)
master_job._commit()
return True
def _resubmit(self):
"""Resubmit a DIRAC job"""
j = self.getJobObject()
parametric = False
script_path = os.path.join(j.getInputWorkspace().getPath(), 'dirac-script.py')
# Check old script
if j.master is None and not os.path.exists(script_path):
raise BackendError('Dirac', 'No "dirac-script.py" found in j.inputdir')
if j.master is not None and not os.path.exists(script_path):
script_path = os.path.join(
j.master.getInputWorkspace().getPath(), 'dirac-script.py')
if not os.path.exists(script_path):
raise BackendError('Dirac', 'No "dirac-script.py" found in j.inputdir or j.master.inputdir')
parametric = True
# Read old script
f = open(script_path, 'r')
script = f.read()
f.close()
# Create new script - ##note instead of using get_parametric_dataset
# could just use j.inputdata.
if parametric is True:
parametric_datasets = get_parametric_datasets(script.split('\n'))
if j.master:
if len(parametric_datasets) != len(j.master.subjobs):
raise BackendError('Dirac', 'number of parametric datasets defined in API script doesn\'t match number of master.subjobs')
if j.inputdata and len(j.inputdata) > 0:
_input_files = [f for f in j.inputdata if not isType(f, DiracFile)]
else:
_input_files = []
if set(parametric_datasets[j.id]).symmetric_difference(set([f.namePattern for f in _input_files])):
raise BackendError(
'Dirac', 'Mismatch between dirac-script and job attributes.')
script = script.replace('.setParametricInputData(%s)' % str(parametric_datasets),
'.setInputData(%s)' % str(parametric_datasets[j.id]))
script = script.replace('%n', str(j.id)) # name
start_user_settings = '# <-- user settings\n'
new_script = script[
:script.find(start_user_settings) + len(start_user_settings)]
job_ident = get_job_ident(script.split('\n'))
for key, value in self.settings.iteritems():
if str(key).startswith('set'):
_key = key[3:]
else:
_key = key
if type(value) is str:
template = '%s.set%s("%s")\n'
else:
template = '%s.set%s(%s)\n'
new_script += template % (job_ident, str(_key), str(value))
new_script += script[script.find('# user settings -->'):]
# Save new script
new_script_filename = os.path.join(j.getInputWorkspace().getPath(), 'dirac-script.py')
f = open(new_script_filename, 'w')
f.write(new_script)
f.flush()
f.close()
return self._common_submit(new_script_filename)
